A copper wire carrying current, a white dwarf the mass of the Sun squeezed into the size of the Earth, a puddle of liquid helium that flows uphill and refuses to boil, the coherent beam of a laser, and a cloud of atoms chilled to a billionth of a degree that suddenly acts like one giant matter-wave — these look like utterly different corners of physics. They are all the same story. Each is a gas of identical quantum particles, cold enough or dense enough that the usual classical picture of "little billiard balls with names" has quietly broken down. What replaces it is not a new force or a new equation of motion. It is a new way of counting.

This page teaches one idea: when identical quantum particles are packed close, the physics is decided by how many of them are allowed to share the same quantum state. Fermions refuse to share; bosons love to pile up. That single yes/no answer splits the quantum world in two and explains why a metal conducts, why a dead star does not collapse, and why lasers and Bose–Einstein condensates exist at all. We will build the two counting rules, meet the occupation-number formulas that encode them, and see exactly when the old classical counting is good enough and when it fails.

When does a gas stop being classical?

In an ordinary room, air molecules are tiny points rattling around with lots of empty space between them. Each molecule is a distinct object you could, in principle, tag and follow. This is the world of the partition function and Maxwell–Boltzmann statistics, and it works beautifully — most of the time.

Quantum mechanics smears each particle out over a length called the thermal de Broglie wavelength,

Cold, light particles are fuzzy (large \lambda); hot, heavy ones are sharp (small \lambda). Now compare that fuzziness with the average spacing between particles, set by the number density n (spacing \sim n^{-1/3}). The whole question of "classical or quantum" rides on one dimensionless number, the degeneracy parameter n\lambda^3:

• n\lambda^3 \ll 1 — the wave-packets barely overlap, particles rarely try to occupy the same state, and the gas is non-degenerate (classical). Room-air sits here by a huge margin, roughly n\lambda^3 \sim 10^{-6}.
• n\lambda^3 \sim 1 — the wave-packets touch and overlap. Now it matters enormously whether two identical particles are allowed to share a state. The gas is degenerate, and quantum statistics takes over. Conduction electrons in a metal are deep in this regime even at room temperature (they are light and dense); helium near 4\ \text{K} reaches it too.

So "quantum gas" is not about being at absurd energies — it is about wave-packets crowding each other until the rule for sharing states becomes the dominant fact of life.

Indistinguishability: particles with no names

Here is the crack in the classical picture. Two electrons are not merely similar — they are identical in the strongest possible sense. There is no scratch, no serial number, no hidden label that tells one from the other. If you swap two of them, you do not get a new arrangement; you get the very same physical state. Nature cannot tell the difference, so neither may our counting.

Classical statistics forgets this. It treats "electron A here, electron B there" and "B here, A there" as two different microstates and counts them both. Gibbs already noticed the resulting over-counting and patched it by dividing the number of arrangements by N! — the number of ways to permute N labels. That 1/N! fix rescues the entropy of a dilute gas (it cures the Gibbs paradox), but it is only an approximation. It silently assumes every particle sits in a different state, so that the permutations are all distinct. The moment two identical particles genuinely compete for the same state — exactly what happens once n\lambda^3 \sim 1 — the naive 1/N! either over- or under-counts, and we need the honest quantum bookkeeping.

Two families: fermions and bosons

Every particle in nature carries an intrinsic spin, and that spin sorts it into one of two camps — with completely opposite social habits.

• Fermions have half-integer spin (\tfrac12, \tfrac32, \dots): electrons, protons, neutrons, and any atom built from an odd number of them. Their combined wavefunction is antisymmetric — it flips sign when you swap two particles. Set two fermions in the same quantum state and that sign flip forces the wavefunction to equal minus itself, hence to vanish: the state is forbidden. This is the Pauli exclusion principle: at most one fermion per quantum state. It is why electrons stack into shells instead of all collapsing into the lowest orbital, and therefore why chemistry, and you, exist.
• Bosons have integer spin (0, 1, 2, \dots): photons, the helium-4 atom, phonons of lattice vibration. Their wavefunction is symmetric — unchanged under a swap. Far from forbidding double occupancy, the symmetry positively encourages it: bosons are gregarious, and any number of them can pile into one and the same state.

The figure above shows the two habits on the same energy ladder. Fermions stack up one per rung, filling every state from the floor up to a top level; bosons abandon the upper rungs and crowd onto the lowest one. Reveal it step by step.

The occupation numbers: one formula, one sign

Statistical mechanics turns those two habits into two tidy formulas for the average number of particles \langle n\rangle occupying a single quantum state of energy \varepsilon, when the gas sits at temperature T with chemical potential \mu. Write the natural combination x = (\varepsilon - \mu)/kT — how far a state's energy sits above the chemical potential, measured in units of the thermal energy. Then:

The two distributions are the same expression, differing only by the sign in the denominator: +1 for fermions, −1 for bosons. The graph plots both against x=(\varepsilon-\mu)/kT, together with the classical Maxwell–Boltzmann curve for comparison.

Two features are worth staring at. Near x = 0 — states within a thermal energy of \mu — the three curves disagree wildly: the fermion curve is politely capped below 1, while the boson curve rockets to infinity. That is the whole difference between refusing to share and loving to share. But out in the tail, at large x, they converge. That convergence is the next, unifying idea.

Why both reduce to the classical law

Look at the denominators. When a state's energy sits several kT above the chemical potential, x is large and e^{x} \gg 1. Against that enormous exponential, the little \pm 1 is a rounding error. Drop it from either formula and both collapse to the same thing:

That limit is precisely the Maxwell–Boltzmann occupation — the classical Boltzmann factor. So Fermi–Dirac and Bose–Einstein are not exotic replacements for classical statistics; they are the full truth, and the classical law is the shadow they both cast in the dilute, high-energy regime n\lambda^3 \ll 1, where states are so sparsely occupied (\langle n\rangle \ll 1) that nobody is ever competing to share. Whether a gas is "quantum" or "classical" is therefore not a property of the particles but of the conditions: warm it up or thin it out and every quantum gas fades gracefully into a classical one.

The Fermi sea and the frozen step

The chemical potential of a fermion gas at absolute zero has a special name: the Fermi energy \varepsilon_F = \mu(T=0). Send T \to 0 in the Fermi–Dirac formula and something dramatic happens. For any state below \varepsilon_F the exponent (\varepsilon-\mu)/kT \to -\infty, so \langle n\rangle \to 1; for any state above it the exponent \to +\infty, so \langle n\rangle \to 0. The smooth curve hardens into a perfect step: every state up to \varepsilon_F is full, every state above is empty. Fermions have poured in and filled the lowest available seats one at a time, right up to the brim. That stack of filled states is the Fermi sea, and its surface is the Fermi energy.

Slide the temperature down in the graph below and watch the step sharpen: at high kT the edge is soft (states near \mu are half-filled), and as kT \to 0 it snaps into a cliff at \varepsilon = \varepsilon_F.

Bosons at low temperature do the opposite and, if anything, stranger thing. With no exclusion rule to stop them, a macroscopic fraction of all the particles avalanches into the single lowest state once the gas is cold enough — a Bose–Einstein condensate, a chunk of matter in which billions of atoms share one quantum state and behave as a single coherent wave. Einstein predicted it in 1924–25; it was finally made in a real gas in 1995. Fermions fill a sea; bosons dive for the floor.

What the counting rule builds

These two ways of filling states are not a bookkeeping curiosity — they hold up stars and light up rooms.

• Degeneracy pressure. Packing fermions tighter forces them into ever-higher unfilled states (the lower ones are taken), and those energetic particles push back hard — a pressure that survives even at T = 0 and owes nothing to heat. Electron degeneracy pressure holds up a white dwarf; when gravity finally wins, neutron degeneracy pressure holds up a neutron star. The same pressure gives metals their stiffness.
• Electrons in a metal. Conduction electrons form a dense degenerate Fermi gas. Only the few near the Fermi surface can absorb energy or carry current, which explains a metal's modest heat capacity and its conductivity — puzzles that classical statistics got badly wrong.
• Photon and phonon gases. Blackbody radiation is a gas of photons — massless bosons with \mu = 0 — and applying Bose–Einstein counting to it yields the Planck spectrum, the birth cry of quantum theory. Lattice vibrations (phonons) are bosons too.
• Coherent quantum matter. The boson urge to share one state underlies the laser (many photons in one mode), the superfluidity of helium-4, Bose–Einstein condensates, and — through paired electrons — the flavour of superconductivity.

Each of these deserves its own page; here the point is only that they all descend from a single question — how many identical particles may share a state? — answered 1 for fermions and \infty for bosons.

Worked examples

Example 1 — occupancy at the chemical potential. Consider a state right at \varepsilon = \mu, so x = 0 and e^{x} = 1. A fermion state has

Exactly half-full — this is why the Fermi energy is often defined as the level of "50% occupancy". A boson state at x=0, by contrast, has \langle n\rangle_{\text{BE}} = 1/(1-1) = \infty: the condensate limit.

Example 2 — a state one thermal energy up. Take x = (\varepsilon-\mu)/kT = 1, so e^{x} = e \approx 2.718. Compare the three counts:

The boson state is the most crowded, the fermion state the least, and classical Maxwell–Boltzmann sits between them — the universal ordering \langle n\rangle_{\text{FD}} < \langle n\rangle_{\text{MB}} < \langle n\rangle_{\text{BE}}.

Example 3 — the classical tail. Now go far out to x = 5, so e^{5} \approx 148.4:

The three answers now agree to about one part in a hundred. Five thermal energies above \mu, the choice of statistics scarcely matters — the promised classical limit, made numerical.